Mixing injection molding process comprehensive emission evaluation modeling method, online dynamic evaluation method and system

Abstract

The invention discloses a mixing injection molding process comprehensive emission evaluation modeling method, an online dynamic evaluation method and a system, and provides a mixing injection molding process-based comprehensive emission evaluation modeling method which mainly comprises a process-emission dynamic coupling modeling method, a comprehensive emission prediction modeling method and a data pre-storage library modeling method. On the basis of an evaluation modeling method, an online dynamic evaluation method is further provided, on the premise that the production process is not interrupted, parameters are collected online, comprehensive emission calculation is completed, comprehensive emission prediction is achieved, and dynamic data updating is completed. The invention further provides an online dynamic evaluation system based on the method. The online dynamic evaluation system comprises a sensing layer, an edge layer and a cloud platform. According to the method, the problems that in a traditional evaluation method, off-line data-based emission accounting is not timely, the accounting type is single, the adaptability is insufficient due to the fact that the influence of process parameters on emission is not fully considered, and the multi-source comprehensive emission dynamic evaluation capability is weak are solved.

Description

TECHNICAL FIELD

[0001] The application relates to the cross field of intelligent manufacturing and green manufacturing, and in particular to a compounding injection molding process comprehensive emission evaluation modeling method, an online dynamic evaluation method and a system. BACKGROUND

[0002] For compounding injection molding process comprehensive emission evaluation, the following problems exist:

[0003] 1) Single-dimensional monitoring: the traditional system only monitors a single emission index or can only obtain carbon emission data through basic calculation, lacks real-time monitoring of pollutants such as VOCs and particulate matter, and cannot capture the comprehensive emissions in the compounding process of multi-phase materials;

[0004] 2) Insufficient dynamic monitoring: existing emission monitoring is mostly based on offline sampling and cannot complete online evaluation of composite emission sources;

[0005] 3) Large static evaluation error: fixed threshold is used for judgment, and the influence of multiple process parameters on comprehensive emissions is not considered, especially for the following specific characteristics of the compounding injection molding process: composite emissions generated by the simultaneous plasticization of multi-phase materials, the coupling influence of dynamic process parameters such as temperature, pressure, and injection rate on emissions, and the inability to obtain a relatively accurate calculation model;

[0006] 4) Lack of systematic solution: traditional solutions are mostly single-link carbon emission accounting or evaluation methods, which are difficult to achieve accurate and complete comprehensive emission analysis when applied to injection molding equipment, lack of comprehensive emission evaluation methods based on compounding injection molding type equipment, and lack of a relatively complete and systematic design method. SUMMARY

[0007] The purpose of the present application is to provide a compounding injection molding process comprehensive emission evaluation modeling method and an online dynamic evaluation method, comprising the following steps:

[0008] 1) Obtain different process parameters and establish a data pre-storage library;

[0009] 2) Use a phase space reconstruction algorithm based on the Takens embedding theorem to reconstruct the process parameters in the data pre-storage library, and construct a joint phase space through tensor product to construct a process-emission dynamic coupling model;

[0010] 3) Based on the process-emission dynamic coupling model, construct a comprehensive emission prediction model based on GA-LSTM;

[0011] 4) Obtain the current process parameters and input the current process parameters into the comprehensive emission prediction model to obtain the comprehensive emission prediction result.

[0012] According to the above two modeling methods, based on the required parameters, a data pre-storage modeling method is proposed as a traceable and accurate instruction-based data library.

[0013] Further, in step 2), the step of constructing the process-emission dynamic coupling model includes:

[0014] 2.1) The phase space of various process parameters is reconstructed to obtain the phase space vector of each process parameter, i.e.

[0015]

[0016]

[0017] where T i (m T ), S i (m S ), E i (m E ), V i (m V ), and pC i (m ρc ) represent the temperature, pressure, energy consumption, injection rate, and emission density phase space vectors, respectively. i , s i , e i , v i , and pC i represent the temperature, pressure, energy consumption, injection rate, and emission density parameters, respectively. T , m S , m E , m V , and m ρc represent the reconstruction dimensions of temperature, pressure, energy consumption, injection rate, and emission density, respectively.

[0018] 2.2) The process phase space vectors are subjected to multivariate synchronous normalization to obtain the process-emission dynamic coupling model X_emission i (m T , m S , m E , m V , and m ρc ), i.e.

[0019] X_emission i (m T , m S , m E , m V , and m ρc ) = [T i (m T ), S i(m S ),E i (m E ),V i (m V ),ρC i (m ρc (6)

[0020] Wherein: T i (m T ) represents the temperature phase space, S i (m S (E) represents the pressure phase space. i (m E ) represents the energy consumption phase space, V i (m V ) represents the injection rate phase space, ρC i (m ρc () represents the emission concentration phase space.

[0021] Furthermore, when the parameter ρC i (m ρc When the three types of emission densities are incorporated, the process-emission dynamic coupling model is as follows:

[0022] X_Comemission i =[X_emission i (Carbon); X_emission i (Vocs); X_emission i (PM2.5)] (7)

[0023] In the formula, X_emission i (Carbon); X_emission i (Vocs); X_emission i (PM2.5) is a dynamic coupling model of process and emission corresponding to carbon, VOCs, and PM2.5.

[0024] Furthermore, in step 3), the steps for constructing the integrated emissions prediction model based on GA-LSTM include:

[0025] 3.1) Noise reduction is performed on the phase space vector of the process parameters to obtain:

[0026]

[0027] In the formula, The noise-reduced phase space vectors are temperature, pressure, energy consumption, injection rate, and emission density; IDWT and DWT represent...

[0028] Inverse discrete wavelet transform and discrete wavelet transform; Tλ is the adaptive threshold; 'db8' is the wavelet base;

[0029] 3.2) On the denoised data Dynamic range compression is performed to obtain:

[0030]

[0031] where, Dynamic range compressed data x t ' = T t ', S t ', E t ', V t ', pc t '; μ ω , σ ω are the mean and standard deviation; ω:t is the length of the sliding window in time steps;

[0032] 3.3) Based on the process phase space reconstruction model and the dynamic range compressed data, an input matrix is constructed, that is:

[0033]

[0034] 3.4) A prediction model is constructed using a genetic algorithm-based LSTM hyperparameter optimization method, that is:

[0035] C pred (t+Δt) = W O ·h t +b o (15)

[0036] where, W O , b o are the weights and biases; h t is the input;

[0037] 3.5) The prediction model is trained using the input matrix and the corresponding comprehensive emission output data, and residual correction is performed during the training process to obtain a comprehensive emission prediction model, that is:

[0038]

[0039] where, λ is the coefficient; is the residual.

[0040] Further, in step 1), the data pre-storage library includes an electrical energy, a material weight, a waste gas collection, a carbon emission, a VOCs gas, a particulate pollutant, a temperature, a pressure, and an injection rate pre-storage library.

[0041] Further, the electrical energy pre-storage library is as follows:

[0042]

[0043] wherein, Electricity(m,n) represents the power data of the mth set of equipment collected for the nth time, e main Electricity_Main represents the power consumption of the main equipment, Electricity_Aux(k) represents the total power consumption of the kth auxiliary equipment; Electricity_Aux(k) = Electricity_Aux(k,1) + Electricity_Aux(k,2) +... + Electricity_Aux(k,n) n Electricity_All(n) represents the total power collected for the nth time, and Electricity_All represents the total power of a batch; Electricity(i,n) represents the power data of the ith set of equipment collected for the nth time;

[0044] The material weight pre-storage repository is as follows:

[0045]

[0046] wherein, Weight(m,n) represents the material weight of the mth set of equipment collected for the nth time; Weight(m,n) = Weight(m,1) + Weight(m,2) +... + Weight(m,n) n Weight_All(n) represents the total weight of the material collected for the nth time;

[0047] The exhaust gas collection pre-storage repository is as follows:

[0048]

[0049] wherein, Gas_all(m,n) represents the amount of exhaust gas collected for the mth set of equipment for the nth time; Gas_all(m,n) = Gas_all(m,1) + Gas_all(m,2) +... + Gas_all(m,n) n Gas_all(m) represents the total amount of exhaust gas collected for the mth set of equipment for the nth time;

[0050] The carbon emission pre-storage repository is as follows:

[0051]

[0052]

[0053] wherein, Carbon(m,n) represents the carbon emission collected for the mth set of equipment for the nth time; G_Dc represents the carbon emission of electric energy and the carbon emission of material; n Carbon_All(n) represents the total carbon emission collected for the nth time; Carbon_Density represents the carbon emission density; Capacity represents the capacity and the total amount of exhaust gas;

[0054] The VOCs gas pre-storage repository is as follows:

[0055]

[0056] wherein, VOCs_Density represents the VOCs gas density; VOCs gas amount; exhaust gas amount; particulate pollutant pre-storage as follows:

[0057]

[0058]

[0059] wherein, particulate pollutant density; particulate pollutant emission amount; temperature pre-storage as follows:

[0060]

[0061] T x＝1 (k) = [t(1), t(2), t(3),..., t(k)] x＝1 (39)

[0062]

[0063] wherein, temperature; yk is an index; k is the number of temperature samples;

[0064] pressure pre-storage as follows:

[0065]

[0066] S x＝1 (k) = [s(1), s(2), s(3),..., s(k)] x＝1 (42)

[0067]

[0068] wherein, pressure;

[0069] injection rate pre-storage as follows:

[0070]

[0071] V x＝1 (k) = [v(1), v(2), v(3),..., v(k)] x＝1 (45)

[0072]

[0073] wherein, injection rate.

[0074] Further, in step 4), after obtaining the comprehensive emission prediction result, the comprehensive emission prediction result is compared with the emission warning threshold, and if the comprehensive emission prediction result is greater than the emission warning threshold, a warning is issued and a traceability search is performed to display the emission data and process parameters corresponding to the period.

[0075] Further, the comprehensive emission prediction result includes emission prediction results of carbon, VOCs gas and particulate matter.

[0076] Further, in step 4), before inputting the current process parameters into the comprehensive emission prediction model, the process parameters are also subjected to noise reduction and dynamic range compression processing.

[0077] A mixing injection molding process comprehensive emission online dynamic evaluation system applying the method, comprising a perception layer, an edge layer and a cloud platform.

[0078] The perception layer is used to collect process parameters.

[0079] The edge layer performs noise reduction and dynamic range compression on the process parameters, and calls the comprehensive emission prediction model to process the noise-reduced and dynamic range-compressed process parameters to obtain a comprehensive emission prediction result.

[0080] The cloud platform is used to train the comprehensive emission prediction model.

[0081] The technical effects of the present application are self-evident, and the beneficial effects of the present application are:

[0082] First, the present application starts from the equipment operation mechanism level, proposes a mixing injection molding process parameter and comprehensive emission coupling modeling method based on the process phase space reconstruction method, further obtains a process-emission coupling function, and helps to solve the problems of insufficient adaptability caused by insufficient consideration of the influence of process parameters on emissions and weak dynamic evaluation capability of multi-source comprehensive emissions.

[0083] Second: the present application proposes a prediction modeling method based on the genetic algorithm LSTM hyperparameter optimization method, further obtains a comprehensive emission prediction function, and can solve the problem of not timely dynamic monitoring in system operation.

[0084] Third: the present application provides an online dynamic evaluation method and system to solve the problems of large static evaluation error, inaccurate evaluation, untimely offline sampling and inability to monitor in real time of the mixing injection molding process. The method and system can complete online data collection, comprehensive emission dynamic evaluation, periodic data updating according to real-time working conditions, and emission process monitoring during process operation.

[0085] Fourth: the data storage library model proposed by the present application makes detailed classification and storage of required parameters and data, which helps to accurately extract batch or single data for calculation, and also can accurately trace the source when the emission is abnormal. Attached Figure Description

[0086] Figure 1 A logical flowchart for constructing the process-emission dynamic coupling model provided by this invention.

[0087] Figure 2 A flowchart illustrating the construction logic of the integrated emission prediction model provided by this invention.

[0088] Figure 3 This invention provides an architecture diagram for an online dynamic assessment system for comprehensive emissions from a mixing and injection molding process.

[0089] Figure 4 The flowchart illustrates the implementation of an online dynamic assessment method for comprehensive emissions from a mixing and injection molding process, as provided by this invention.

[0090] Figure 5 This is a schematic diagram of the implementation process for edge layer data preprocessing and pre-storage provided by the present invention.

[0091] Figure 6 The logical flowchart for online integrated emissions assessment provided by this invention. Detailed Implementation

[0092] The present invention will be further described below with reference to embodiments, but it should not be construed that the scope of the present invention is limited to the following embodiments. Various substitutions and modifications made based on ordinary technical knowledge and common practices in the art without departing from the above-described technical concept of the present invention should be included within the scope of protection of the present invention.

[0093] Example 1:

[0094] See Figures 1 to 6 A comprehensive emission assessment modeling method and online dynamic assessment method for compounding injection molding process, comprising the following steps:

[0095] 1) Obtain different process parameters and establish a pre-repository data library;

[0096] 2) The process parameters in the data pre-store are reconstructed using a phase space reconstruction algorithm based on Takens' embedding theorem, and a joint phase space is constructed through tensor product, thereby constructing a dynamic coupling model of process and emission.

[0097] 3) Based on the process-emission dynamic coupling model, a comprehensive emission prediction model based on GA-LSTM is constructed;

[0098] 4) Obtain the current process parameters and input them into the integrated emission prediction model to obtain the integrated emission prediction results.

[0099] Based on the two modeling methods mentioned above and their required parameters, a data pre-repository modeling method is proposed as a traceable data foundation that can be extracted according to accurate instructions.

[0100] Example 2:

[0101] A comprehensive emission assessment modeling method and online dynamic assessment method for a compounding injection molding process, with the same technical content as in Example 1, further comprising, in step 2), the step of constructing the process-emission dynamic coupling model includes:

[0102] 2.1) Reconstruct the phase space of various process parameters to obtain the phase space vectors of each process parameter, i.e.:

[0103]

[0104] In the formula, T i (m T ), S i (m S E i (m E V i (m V ), ρC i (m ρc ) represent the phase space vectors of temperature, pressure, energy consumption, injection rate, and emission density, respectively; t i s i e i v i ,ρc i They represent temperature respectively

[0105] Temperature, pressure, energy consumption, injection rate, and emission density parameters; m T m S m E m V m ρc respectively

[0106] Represents the reconstructed dimension of temperature, pressure, energy consumption, injection rate, and emission density;

[0107] 2.2) Perform multivariate synchronous normalization on the phase space vectors of each process to obtain the process-emission dynamics.

[0108] State coupling model X_emission i (m T ,m S ,m E ,m V ,m ρc ),Right now:

[0109] X_emission i (m Tm S ,m E ,m V ,m ρc )=[T i (m T ),S i (m S ),E i (m E ),V i (m V ),ρC i (m ρc )] (6)

[0111] Wherein: T i (m T ) is the temperature phase space, S i (m S ) is the pressure phase space, E i (m E ) is the energy consumption phase space

[0112] , V i (m V ) is the injection rate phase space, ρC i (m ρc ) is the emission concentration phase space,

[0113] Example 3:

[0114] A mixing injection molding process comprehensive emission evaluation modeling method and online dynamic evaluation method, the technical content is the same as any one of embodiments 1-2, further, when the parameter ρC i (m ρc ) is brought into three kinds of emission density, the process-emission dynamic coupling model is as follows:

[0115] X_Comemission i =[X_emission i (Carbon);X_emission i (Vocs);X_emission i (PM2.5)] (7)

[0117] In the formula, X_emission i (Carbon);X_emission i (Vocs);X_emission i (PM2.5) is

[0118] The process-emission dynamic coupling model corresponding to carbon, VOCs, PM2.5.

[0119] Example 4:

[0120] A comprehensive emission assessment modeling method and an online dynamic assessment method for a compounding injection molding process are provided. The technical content is the same as any one of Examples 1-3. Further, in step 3), a comprehensive emission prediction method based on GA-LSTM is constructed.

[0121] The steps for testing the model include:

[0122] 3.1) Noise reduction is performed on the phase space vector of the process parameters to obtain:

[0123]

[0124] In the formula, The noise-reduced phase space vectors are temperature, pressure, energy consumption, injection rate, and emission density; IDWT and DWT represent...

[0125] Inverse discrete wavelet transform and discrete wavelet transform; T λ 'db8' is an adaptive threshold; 'db8' is a wavelet basis.

[0126] 3.2) For the denoised data Dynamic range compression yields:

[0127]

[0128] In the formula, Data x after dynamic range compression t ' ＝ T t ',S t ',E t ',V t ',ρc t ';μ ω σ ω ω:t represents the mean and standard deviation; ω:t is the length of the sliding window in units of time steps.

[0129] 3.3) Based on the process phase space reconstruction model and the dynamically compressed data, construct the input matrix, i.e.:

[0130]

[0131] 3.4) Construct a prediction model using the LSTM hyperparameter optimization method based on genetic algorithms, i.e.:

[0132] C pred (t+Δt)=W O ·h t +b o (15)

[0133] In the formula, WO , b o are weights and biases; h t is input;

[0134] 3.5) Train the prediction model with the input matrix and the corresponding comprehensive emission output data, and make residual correction during the training process to obtain a comprehensive emission prediction model, i.e.

[0135]

[0136] wherein, λ is a coefficient; is residual error.

[0137] Example 5:

[0138] A compounding injection molding process comprehensive emission evaluation modeling method and online dynamic evaluation method, the technical content is the same as any one of embodiments 1-4, further, in step 1), the data pre-storage library includes electric energy, material weight, waste gas collection, carbon emission, VOCs gas, particulate pollutants, temperature, pressure, injection rate pre-storage library.

[0139] Example 6:

[0140] A compounding injection molding process comprehensive emission evaluation modeling method and online dynamic evaluation method, the technical content is the same as any one of embodiments 1-5, further, the electric energy pre-storage library is as follows:

[0141]

[0142]

[0143] wherein, represents the electric power data collected for the mth set of equipment for the nth time, e main represents the electric power consumed by the main equipment, represents the total electric power consumed by the k auxiliary equipment; Electricity n represents the total electric power collected for the nth time, and Electricity_All represents the total electric power of a batch; represents the electric power data of the ith set of equipment collected for the nth time;

[0144] The material weight pre-storage library is as follows:

[0145]

[0146] wherein, is the material weight of the rth set of equipment collected for the nth time; Weight(j) n is the total weight of the material collected for the nth time;

[0147] The waste gas collection pre-storage library is as follows:

[0148]

[0149] wherein, Gas_all is the total amount of exhaust gas collected for the mth set of equipment; Gas_all n Gas_all is the total amount of exhaust gas collected for the mth set of equipment; Gas_all

[0150] Carbon emission pre-repository is as follows:

[0151]

[0152]

[0153] wherein, G_Dc is the carbon emission for the mth set of equipment collected for the nth time; G_Dc is the carbon emission for the mth set of equipment collected for the nth time; n G_Dc is the carbon emission for the mth set of equipment collected for the nth time; G_Dc is the carbon emission for the mth set of equipment collected for the nth time; G_Dc is the carbon emission for the mth set of equipment collected for the nth time;

[0154] VOCs gas pre-repository is as follows:

[0155]

[0156] wherein, G_Dc is the carbon emission for the mth set of equipment collected for the nth time; G_Dc is the carbon emission for the mth set of equipment collected for the nth time; G_Dc is the carbon emission for the mth set of equipment collected for the nth time;

[0157]

[0158] wherein, G_Dc is the carbon emission for the mth set of equipment collected for the nth time; G_Dc is the carbon emission for the mth set of equipment collected for the nth time; Temperature pre-repository is as follows:

[0159]

[0160] T x＝1 (k) = [t(1), t(2), t(3),..., t(k)] x＝1 (39)

[0161]

[0162] wherein, T is the temperature; yk is the index; k is the number of temperature samples;

[0163] Pressure pre-repository is as follows:

[0164]

[0165] S x＝1 (k) = [s(1), s(2), s(3),..., s(k)] x＝1 (42)

[0166]

[0167] wherein, is the pressure;

[0168] The injection rate pre-stored repository is as follows:

[0169]

[0170] V x＝1 (k) = [v(1), v(2), v(3),..., v(k)] x＝1 (45)

[0171]

[0172] wherein, is the injection rate.

[0173] Example 7:

[0174] A compounding injection molding process comprehensive emission evaluation modeling method, online dynamic evaluation method, the technical content is same as any one of embodiments 1-6, further, in step 4), after obtaining the comprehensive emission prediction result, the comprehensive emission prediction result is compared with the emission warning threshold, if the comprehensive emission prediction result is greater than the emission warning threshold, a warning is issued and the corresponding emission data and process parameters of the period are displayed by tracing back and searching.

[0175] Example 8:

[0176] A compounding injection molding process comprehensive emission evaluation modeling method, online dynamic evaluation method, the technical content is same as any one of embodiments 1-7, further, the comprehensive emission prediction result includes the emission prediction result of carbon, VOCs gas and particulate matter.

[0177] Example 9:

[0178] A compounding injection molding process comprehensive emission evaluation modeling method, online dynamic evaluation method, the technical content is same as any one of embodiments 1-8, further, in step 4), before inputting the current process parameters into the comprehensive emission prediction model, the process parameters are also subjected to noise reduction and dynamic range compression processing.

[0179] Example 10:

[0180] A mixing injection molding process comprehensive emission online dynamic evaluation system applying the method of any one of embodiments 1-9, comprising: a perception layer, an edge layer, and a cloud platform;

[0181] The perception layer is used to collect process parameters;

[0182] The edge layer performs noise reduction and dynamic range compression on the process parameters, and calls a comprehensive emission prediction model to process the noise-reduced and dynamic range-compressed process parameters to obtain a comprehensive emission prediction result;

[0183] The cloud platform is used to train the comprehensive emission prediction model.

[0184] Embodiment 11:

[0185] In a mixing injection molding process comprehensive emission evaluation modeling method, the core three types of model construction methods include: step S101: data preprocessing and storage repository construction; step S102: construction of process-emission dynamic coupling model and function obtaining method; and step S103: construction of comprehensive emission prediction model and function obtaining method.

[0186] Step S101: data pre-storage repository modeling

[0187] It mainly includes orderly naming the data collected by each group and storing it in the corresponding model matrix for later calculation or traceability. Specifically, it includes:

[0188] Electricity includes main equipment electricity + auxiliary equipment electricity:

[0189] Wherein Electricity represents the electric power data collected for the mth set of equipment for the nth time, e main Electricity represents the electric power consumed by the main equipment, Electricity represents the total electric power consumed by the k auxiliary equipment.

[0190] The data storage model is:

[0191]

[0192] Wherein Electricity represents the electric power data collected for the mth set of equipment for the nth time, Electricity n Electricity represents the total electric power collected for the nth time, and Electricity_All represents the total electric power of a batch.

[0193] In the production of a batch of products, the weight of each type of raw material is generally determined according to the corresponding proportion and weight of the scheme, and the feeding, melting mixing and extruder conveying part are continuously working, and the weight of the raw material used for each injection molding product cannot be specifically determined, therefore, the pre-processing and storage of the weight of the raw material are carried out in a batch unit.

[0194] The weight of the raw material can be obtained by a weighing sensor, Indicates the n number of parts, the raw material quality of the r batch, and the coefficient 1 indicates the raw material.

[0195] The variables W(1) and Weight(1) are stored, and the data storage model is:

[0196]

[0197] Weight(1) n Indicates the total amount of raw materials for injection molding of the n number of parts in r batches.

[0198] In the product output stage, the product quality can be obtained by a weighing sensor, and the Indicates the weight of the product obtained by injection molding of the n number of parts for the m time, and the coefficient 2 indicates the product.

[0199] The data storage model is:

[0200]

[0201] Weight(2) n Indicates the total amount of products for injection molding of the n number of parts in a batch.

[0202] For the online mixing and injection equipment, the exhaust emission points are mainly concentrated on the two exhaust interfaces of the double screw extruder conveying channel, and the vacuum device and interface are arranged in the injection section, so that the electronic exhaust collection box is installed at the exhaust emission point, and the Indicates the amount of exhaust gas generated by injection molding of the n number of parts for the m time, and the average value is obtained after multiple collection and screening to delete the extreme value, and the patent discusses a scheme of rapid and continuous five times analysis.

[0203] GL(5) = [gl(1), gl(2), gl(3), gl(4), gl(5)]

[0204] GL(5) max = MAX[GL(5)]; GL(5) min = min[GL(5)]

[0205] The set after removing the maximum and minimum values is represented as: GL(3) = [gl(x1), gl(x2), gl(x3)]

[0206]

[0207] Data storage model is:

[0208]

[0209] Where, Gas_all n Indicates the amount of all exhaust gas generated by one batch injection molding of n number of parts.

[0210] Carbon emission pre-repository:

[0211]

[0212] The capacity is calculated by the carbon emission of electric energy and the carbon emission of materials, and the density Dc is calculated by the capacity ÷ total exhaust gas amount. Here it is different from VOCs and particulate matter, which is measured density, multiplied by the total exhaust gas amount to calculate the capacity.

[0213] Similarly, based on metal oxide VOCs gas sensor. Stored in set Dv(5), using Indicates the VOCs concentration generated by the n number of parts, the mth injection molding, the model is:

[0214] Dv(5) = [dv(1), dv(2), dv(3), dv(4), dv(5)]

[0215] Dv(5) max = MAX[Dv(5)]; Dv(5) min = min[Dv(5)]

[0216] The set after removing the maximum and minimum values is represented as: Dv(3) = [dv(x1), dv(x2), dv(x3)]

[0217]

[0218] Store VOCs concentration with Dv and content with G_Dv, data storage model is:

[0219]

[0220] Where, G_Dv(i,j) indicates the amount of VOCs generated by the i number of parts, the jth injection molding, G_Dv n Indicates the amount of all VOCs generated by one batch injection molding of n number of parts.

[0221] Similarly, set particulate matter analysis instrument at the exhaust emission point, stored in set Dpm(5), using Indicates the particulate pollutant concentration generated by the n number of parts, the mth injection molding, the model is:

[0222] Dpm(5) = [dpm(1), dpm(2), dpm(3), dpm(4), dpm(5)]

[0223] Dpm(5) max = MAX[Dpm(5)]; Dpm(5) min = min[Dpm(5)]

[0224] The set after removing the maximum and minimum values is represented as: Dpm(3) = [dpm(x1), dpm(x2), dpm(x3)]

[0225]

[0226] The particle density is stored using the function Dpm, and the content is stored using the function G_Dpm, and the storage model is:

[0227]

[0228]

[0229] Where G_Dpm(i,j) represents the amount of particle contamination generated by the i-th part during the j-th injection molding, and G_Dpm n represents the total amount of particle contamination generated by the n-th part during this batch of injection molding.

[0230] Similarly, temperature sensors are set in the material preparation bin, material melting and mixing section, screw extrusion section, pressure maintaining section, and injection molding section of the dryer. Similarly, the temperature of each part is collected multiple times quickly, and the maximum and minimum values are filtered and deleted, and the average value is obtained, which is represented as represents the temperature of the x section during the m-th injection molding process of the n-th part. Since the present application needs to collect the temperatures of five parts, x ∈ [1, 5] and x is an integer, and the parameter x corresponds to Table 1.

[0231] Table 1 Temperature parameter correspondence diagram

[0232]

[0233] Taking x = 1 as an example, the temperature function is:

[0234] T x＝1 (5) = [t(1), t(2), t(3), t(4), t(5)] x＝1

[0235] T x＝1 (5) max = MAX[T x＝1 (5)]; T x＝1 (5) min = min[Tx＝1 (5)]

[0236] The set after removing the maximum and minimum values is represented as: T x＝1 (3) = [t(y1), t(y2), t(y3)] x＝1

[0237]

[0238] The data storage model is:

[0239]

[0240] Similarly, pressure sensors are deployed in the screw extrusion section, pressure maintaining section, and injection section, represents the pressure of part x of the n-th part in the m-th injection process, x ∈ [1, 3] and x is an integer. The parameter x corresponds to Table 2.

[0241] Table 2: Correspondence of pressure parameters

[0242]

[0243] Taking x = 1 as an example, the pressure function is:

[0244] S x＝1 (5) = [s(1), s(2), s(3), s(4), s(5)] x＝1

[0245] S x＝1 (5) max = MAX[S x＝1 (5)]; S x＝1 (5) min = min[S x＝1 (5)]

[0246] The set after removing the maximum and minimum values is represented as: S x＝1 (3) = [s(y1), s(y2), s(y3)] x＝1

[0247]

[0248] The data storage model is:

[0249]

[0250] Similarly, rate sensors are deployed in the screw extrusion section and the injection cylinder, represents the fluid rate of part x of the n-th part in the m-th injection process, x ∈ [1, 2] and x is an integer, and parameter x = 1 corresponds to the screw extrusion section, and x = 2 corresponds to the injection cylinder.

[0251] For example, when x = 1, the rate function is:

[0252] V x＝1 (5) = [v(1), v(2), v(3), v(4), v(5)] x＝1

[0253] V x＝1 (5) max = MAX[V x＝1 (5)];V x＝1 (5) min = min[V x＝1 (5)]

[0254] The set after removing the maximum and minimum values is represented as: V x＝1 (3) = [v(y1), v(y2), v(y3)] x＝1

[0255]

[0256] The data storage model is:

[0257]

[0258] Further calculation of carbon emission data, according to the carbon accounting function, carbon emission calculation of electric energy consumption, carbon emission of enterprise electricity = total electricity × grid emission factor - carbon emission reduction represented by green certificate, which can be expressed as: C E = Electricity × ef - C c .

[0259] Based on material data, in the case of material thermal decomposition, material balance is used to reverse carbon emission, which is specifically:

[0260] Carbon emission = (carbon content of raw material - carbon content of product) * carbon conversion coefficient * global warming trend

[0261] Which can be expressed as:

[0262] Where: Q(1) i represents the carbon content ratio of the i-th raw material, and Q(2) represents the carbon content ratio of the product.

[0263] Step S102: Construction of process-emission dynamic coupling model and function acquisition

[0264] Please refer to Figure 1 The multi-element process parameters in the mixing and injection molding process of thermoplastic composite materials, such as working temperature, pressure, injection rate, and emission concentration, are fused and reconstructed to build a process-emission dynamic coupling model.

[0265] The phase space reconstruction method based on Takens' embedding theorem corresponds to the following model:

[0266] Let the original process parameter sequence be: {x i The reconstructed phase space vector is: |i=1,2,...,N}.

[0267] X i (m)=[x i ,x (i+τ) ,x (i+2τ) ,...x (i+(m-1)τ) ]

[0268] Where: m is the embedding dimension, which determines the degrees of freedom of the reconstructed system; τ is the time delay (time delay parameter), which avoids excessive correlation between adjacent components.

[0269] According to the phase space reconstruction method, this invention proposes that, firstly, various process parameters are input, and after each parameter is reconstructed independently, a joint phase space is constructed through tensor product.

[0270] The reconstructed phase space vectors of each parameter are:

[0271]

[0272]

[0273] Furthermore, the reconstructed phase space vectors are subjected to multivariate synchronous normalization, and the model is as follows:

[0274] X_emission i (m T ,m S ,m E ,m V ,m ρc )=[T i (m T ),S i (m S ),E i (m E ),V i (m V ),ρC i (m ρc )]

[0275] Wherein: T i (m T ) represents the temperature phase space, S i (m S (E) represents the pressure phase space. i (m E (Energy consumption phase space)

[0276] Between, V i (m V ) represents the injection rate phase space, ρC i (m ρc () represents the emission concentration phase space.

[0277] Specifically, the parameter ρC i (m ρc The emission density can be categorized into three types, and the carbon emission concentration can be categorized accordingly. After phase space coupling, X_emission i This refers to a model that incorporates the effects of process parameters on carbon emissions, particulate matter emissions, and VOCs emissions.

[0278] Integrating these three elements yields a complete process-emission dynamic coupling function, which is the comprehensive emission calculation function:

[0279] X_Comemission i =[X_emission i (Carbon); X_emission i (Vocs); X_emission i (PM2.5)

[0280] Step S103: Construction of the integrated emission prediction model and acquisition of functions

[0281] Please see Figure 2 Based on the above process-emission dynamic coupling modeling, a GA-LSTM model is further constructed.

[0282] The process involves completing the predictive modeling, specifically including: data denoising, predictive model construction, and residual correction.

[0283] The data denoising includes: using wavelet thresholding to denoise the original data to reduce sensor drift interference.

[0284] Disturbance. That is, to make the process parameters of this invention phase space heterogeneous function T... i (m T ), S i (m S E i (m E ),

[0285] V i (m V ), ρC i (m ρc Substituting each into the denoising formula, we get:

[0286]

[0287] The adaptive threshold T λ for:

[0288]

[0289] Among them, C HH σ represents the highest frequency wavelet coefficients, and σ is the noise standard deviation estimate.

[0290] The above function values ​​can be abbreviated as follows:

[0291] For the denoised data Dynamic range compression is performed using the following formula:

[0292]

[0293] Will Substitute x into each t We can obtain: T t ',S t ',E t ',V t ',ρc t '.

[0294] The prediction model construction includes: based on the process phase space reconstruction model and the denoised data, inputting the T... t ',S t ',E t ',V t ',ρc t ', further construct the input matrix, the matrix model is:

[0295]

[0296] T, S, E, V, and ρc represent temperature, pressure, energy consumption, and emission concentration, respectively.

[0297] Further optimization of hyperparameters using a genetic algorithm yields the following prediction formula model:

[0298] C pred (t+Δt)=W O ·h t +b o

[0299] The residual correction is mainly used to address the problems of long-term prediction error accumulation and lag in response to sudden process changes. The final corrected output comprehensive emission prediction function is:

[0300]

[0301] Where C final (t+Δt) represents the corrected predicted emission concentration, with weight λ=ρ k (ρ=0.85)

[0302] k is the prediction step length.

[0303] The above is a detailed description of an embodiment of the mixing injection molding process comprehensive emission evaluation modeling method provided by the present application, and it includes the obtaining method of the comprehensive emission calculation function and the comprehensive emission prediction function. The following is a detailed description of a mixing injection molding process comprehensive emission online dynamic evaluation method and system provided by the present application on the basis of the above embodiment.

[0304] Please refer to Figure 3 , a mixing injection molding process comprehensive emission online dynamic evaluation system architecture diagram is provided by the present application. The system architecture serves as the system of the evaluation method of the present application, and helps to more clearly illustrate the specific implementation steps of the mixing injection molding process comprehensive emission online dynamic evaluation provided by the present application.

[0305] Further, please refer to Figure 4 , the mixing injection comprehensive emission online dynamic evaluation method provided by the present application mainly includes: S201: online acquisition of multi-source data in the perception layer; S202: data preprocessing and storage in the edge layer; S203: comprehensive emission calculation in the edge layer; S204: comprehensive emission prediction in the edge layer; S205: dynamic optimization evaluation function of the cloud platform.

[0306] Step S201: online acquisition of multi-source data in the perception layer

[0307] The deployment of the multi-source data acquisition sensor array is shown in Table 3.

[0308] Table 3 Deployment of sensor array

[0309]

[0310] The implementation method in this part mainly includes:

[0311] When the working equipment is started, the system is started synchronously, and based on the data synchronization principle, online data acquisition is completed under the condition that the equipment is continuously running, wherein the data to be collected includes: electric energy, raw material weight, product weight, injection rate, pressure, temperature, VOCs emission concentration, particulate matter emission concentration, and total exhaust gas.

[0312] Step S202: data preprocessing and storage in the edge layer

[0313] Please refer to Figure 5 , the implementation method of multi-source data preprocessing and storage, specifically includes:

[0314] First, the validity of the collected data is tested, if the data is invalid, the database is returned, the edge output signal is given to the perception layer to re-collect, if the data is valid, the pre-processing of each part of the data is completed according to the pre-processing rules described in the application content, and further, the data storage is completed according to the storage library model described above.

[0315] The multi-source data mainly includes: electric energy, material weight, waste gas collection, carbon emission, VOCs gas, particulate pollutants, temperature, pressure, injection rate.

[0316] Step S203: edge layer comprehensive emission online dynamic evaluation

[0317] Please refer to Figure 6 The application provides an implementation method based on comprehensive emission online dynamic evaluation.

[0318] When the working equipment starts, the system is started synchronously, the initialization is completed, the periodic cycle is started, and the command is waited for If no command is received, when the timing period arrives, the data collection is completed, the comprehensive emission data is calculated, the prediction is further completed, the data is stored in the historical database, and the counter is cleared to start the next cycle.

[0319] When receiving the real-time query command of the system, the command type is judged, if it is judged that the real-time emission is queried, the current data is collected to complete the calculation, and the current comprehensive emission data is further output. If it is judged that the historical data is queried, the historical data pointed by the command is searched, and the query result is output. After completing the command task, it jumps back to the main line task cycle and waits for the next command.

[0320] The system circulates, so the comprehensive emission data will follow the equipment operation to realize periodic dynamic update, and the related historical data can be traced back, and the whole emission evaluation process is synchronized in the injection molding equipment operation process. It should be noted that the period setting here can be adjusted according to the actual situation, such as several hours, half a day, one day, or one batch.

[0321] Step S204: cloud platform dynamic optimization evaluation function

[0322] With the operation of the equipment and the system, the central processing system of the cloud platform periodically acquires new process parameters and emission data from the historical database, based on the process-emission dynamic coupling model described in step S101, the comprehensive emission evaluation function is trained and optimized in a cycle;

[0323] Based on the comprehensive emission prediction model described in step S102, the comprehensive emission evaluation function is trained and optimized in a cycle;

[0324] Further, the optimized evaluation function and prediction function are transmitted back to the edge layer to complete function update, which is used for analysis and calculation of the edge layer next time.

[0325] Embodiment 13:

[0326] Firstly, when receiving the signal of the control terminal, the sensor array carries out online acquisition of multi-source data, mainly including process parameters such as electric energy, temperature, pressure, and weight data of materials and products;

[0327] According to the aforementioned data pre-storage modeling method, data preprocessing is carried out, and then the data is stored in the warehouse;

[0328] According to the aforementioned process-emission dynamic coupling calculation function, the aforementioned required data is brought in to complete the calculation of comprehensive emissions, and the emission data can be transmitted to the control terminal or updated to the visualization interface and other devices;

[0329] According to the aforementioned comprehensive emission prediction function, the prediction of comprehensive emissions is completed, and the prediction data can be transmitted to the control terminal or updated to the visualization interface and other devices;

[0330] The prediction result is compared with the emission warning threshold, and when the prediction result is higher than the threshold, a warning is issued and the source is traced back to retrieve and display the emission data and process parameters corresponding to the period, so as to quickly find and solve the device exception;

[0331] According to the evaluation result, the historical data is stored, and the counter is further cleared, and the next cycle is started;

[0332] Finally, the new round of parameter acquisition and comprehensive emission calculation data are further brought into the process-emission dynamic coupling model and the comprehensive emission prediction model, and the optimization of the model and the optimization of the evaluation function are completed based on deep learning training, the function is updated, and the cycle is repeated. Therefore, the system comprehensive emission data will follow the device operation to realize periodic dynamic update, and the related historical data can be traced back, and the entire emission evaluation process is carried out synchronously during the operation of the injection molding device.

[0333] It should be noted that the evaluation function specifically includes the process-emission dynamic coupling function and the comprehensive emission prediction function.

[0334] It should be noted that the comprehensive emission data here includes three aspects, namely carbon emission data, VOCs gas emission data, and particulate matter emission data.

[0335] It should be noted that the threshold setting of the prediction part can be set according to the working experience of the specific device, or according to the regional emission requirements. Further adjustment procedures can be added after the warning, such as adjusting the injection molding rate, production rhythm and other process parameters, to further realize the optimization control of injection molding production.

[0336] According to the mixing injection molding process comprehensive emission evaluation modeling method and the online dynamic evaluation method, a mixing injection molding process comprehensive emission online dynamic evaluation system is further provided, which specifically comprises a perception layer, an edge layer and a cloud platform.

[0337] The perception layer is mainly responsible for early data collection, and the architecture specifically comprises a multi-source perception array and data collection.

[0338] The edge layer mainly completes data processing, comprehensive emission calculation, prediction, and data updating, and the architecture specifically comprises an evaluation unit and a preprocessing and prestorage unit.

[0339] The cloud platform is mainly responsible for model training, function optimization, and connection to an upper computer, and the architecture specifically comprises a central processing system comprising a dynamic model training module, a historical database, and a visualization interface.

[0340] The system operation rules include:

[0341] The perception layer receives a control signal, performs online multi-source data collection, and transmits process parameters such as electric energy, temperature, and pressure, and material weight data to the edge layer.

[0342] The edge layer first performs data preprocessing and storage, further performs comprehensive emission dynamic evaluation based on the process-emission dynamic coupling function fed back by the cloud platform, and outputs comprehensive emission data.

[0343] The edge layer performs comprehensive dynamic emission prediction based on the comprehensive emission prediction function fed back by the cloud platform. It should be noted that the comprehensive emission data includes three aspects, namely carbon emission data, VOCs gas emission data, and particulate matter emission data.

[0344] The cloud platform periodically updates the comprehensive emission evaluation model and the comprehensive emission prediction model based on historical database data, and further feeds back the process-emission dynamic coupling function and the comprehensive emission prediction function output after updating the model to the edge layer for dynamic optimization.

[0345] Specifically, the model training includes the establishment and updating of the process-emission dynamic coupling model and the establishment and updating of the comprehensive emission prediction model.

[0346] The function optimization includes optimizing the process-emission dynamic coupling model and the comprehensive emission prediction model, and synchronously updating the process-emission dynamic coupling function and the comprehensive emission prediction function.

[0347] The multi-source perception array in the perception layer comprises an energy perception system, a process parameter perception system, and a pollution perception system.

[0348] The energy perception system comprises a smart electric meter and a raw material and product weighing module.

[0349] The process parameter sensing system includes an extruder rate sensor, an injection rate sensor, an injection pressure sensor, and an injection temperature sensor.

[0350] The pollution sensing system includes a metal oxide VOCs sensor for monitoring the concentration of VOCs gas and a multi-spectrum particulate analyzer for detecting the concentration of particulate matter in the exhaust gas.

[0351] The data collection mainly includes a collection method and a collection target.

[0352] The collection method is as follows: in order to ensure the effectiveness and stability of the collected data, in addition to the collection of power and weight, for other sensor modules, the present application adopts a method of setting a short-time multiple value for each collection command, and then filtering and deleting the extreme values to obtain the average value, thereby obtaining parameters with higher reliability.

[0353] The collection target includes:

[0354] Energy consumption: the device uses electric energy E in aspects of heating and heat preservation, transportation, driving motor, and driving hydraulic pump.

[0355] Material weight: the raw material weight W1 and the product weight W2 are obtained through the weighing module.

[0356] Pollutant emission: the total amount of exhaust gas emission G is directly monitored.

[0357] Pollutant classification: the specific components and concentration ρC of the exhaust gas are obtained from the VOCs sensor, the multi-spectrum particulate analyzer, and the flue gas analyzer.

[0358] Basic process parameters: the temperature of each part is represented by a matrix T, the pressure of each part is represented by a matrix S, and the injection rate of each part is represented by a matrix V.

[0359] According to the data collection, data preprocessing and storage are further completed.

[0360] According to the electric energy data, indirect carbon emissions are calculated, and according to the material and product weight, in the case of material thermal decomposition, material balance is used to deduce carbon emissions. The carbon emissions here do not consider the influence of process parameters, and the results are not accurate, and are only used as calculation parameters for subsequent calculation of comprehensive emissions in the process-emission coupling function.

[0361] Based on the above rules, combined with the aforementioned online dynamic evaluation method of comprehensive emissions, a complete online dynamic evaluation system is formed.

Claims

1. A compounding injection molding process integrated emission assessment modeling method, online dynamic assessment method, characterized in that, The method comprises the following steps: 1) obtaining different process parameters and establishing a data pre-storage library; 2) reconstructing the process parameters in the data pre-storage library using a phase space reconstruction algorithm based on the Takens embedding theorem, and constructing a joint phase space through tensor product to construct a process-emission dynamic coupling model; 3) based on the process-emission dynamic coupling model, constructing a comprehensive emission prediction model based on GA-LSTM; 4) obtaining the current process parameters and inputting the current process parameters into the comprehensive emission prediction model to obtain the comprehensive emission prediction result.

2. The modeling method of claim 1, wherein, In step 2), the step of constructing the process-emission dynamic coupling model comprises: 2.1) reconstructing the phase space of each type of process parameter to obtain the phase space vector of each process parameter, i.e. where T i (m T ), S i (m S ), E i (m E ), V i (m V ), and pC i (m ρc ) represent temperature, pressure, energy consumption, injection rate, and emission density phase space vectors, respectively; t i , s i , e i , v i , and pC i represent temperature, pressure, energy consumption, injection rate, and emission density parameters, respectively; m T , m S , m E , m V , and m ρc represent the reconstruction dimensions of temperature, pressure, energy consumption, injection rate, and emission density, respectively. 2.2) Multivariate synchronously normalizing the process phase space vectors to obtain a process-emission dynamic coupling model X_emission i (m T ,m S ,m E ,m V ,m ρc ), i.e. X_emission i (m T ,m S ,m E ,m V ,m ρc ) = [T i (m T ), S i (m S ), E i (m E ), V i (m V ), pC i (m ρc )] (6) where: T i (m T ) is the temperature phase space, S i (m S ) is the pressure phase space, E i (m E ) is the energy consumption phase space, V i (m V ) is the injection speed phase space, pC i (m ρc ) is the emissions concentration phase space.

3. The modeling method of claim 2, wherein, When the parameter pC i (m ρc ) is substituted into the three types of emission densities, the process-emission dynamic coupling model is as follows: X_Comemission i = [X_emission i (Carbon); X_emission i (Vocs); X_emission i (PM2.5)] (7) In the formula, X_emission i (Carbon) X_emission i (Vocs) X_emission i (PM2.5) is a process-emission dynamic coupling model corresponding to carbon, VOCs, PM2.

5.

4. The modeling method of claim 1, wherein, In step 3), the step of constructing the comprehensive emission prediction model based on GA-LSTM comprises: 3.1) denoising the process parameter phase space vector to obtain: wherein are the temperature, pressure, energy consumption, injection rate, and emissions density phase space vectors after noise reduction; IDWT and DWT represent inverse discrete wavelet transform and discrete wavelet transform, respectively; T λ is the adaptive threshold; 'db8' is the wavelet basis. 3.2) On the de-noised data Dynamic range compression is performed, resulting in: wherein data x after dynamic range compression t ' = T t ', S t ', E t ', V t ', pc t '; μ ω , σ ω are the mean and standard deviation; ω:t is the length of the sliding window in time steps. 3.3) constructing an input matrix based on the process phase space reconstruction model and the data after dynamic range compression, i.e. 3.4) constructing a prediction model using a genetic algorithm-based LSTM hyperparameter optimization method, i.e. C pred (t+Δt) = W O ·h t +b o (15) where W O , b o are weights and biases; h t is the input; 3.5) training the prediction model using the input matrix and the corresponding comprehensive emission output data, and performing residual correction during the training process to obtain the comprehensive emission prediction model, i.e. where λ is a coefficient; is a residual.

5. The modeling method of claim 1, wherein, In step 1), the data pre-storage library comprises an electrical energy, material weight, exhaust gas collection, carbon emission, VOCs gas, particulate pollutant, temperature, pressure, and injection rate pre-storage library.

6. The modeling method of claim 5, wherein, The electrical energy pre-storage library is as follows: In the formula, represents the power data collected for the mth set of equipment for the nth time, e main represents the power consumed by the main equipment, represents the total power consumed by the k auxiliary equipment; Electricity n represents the total power collected for the nth time, and Electricity_All represents the total power of a batch; represents the power data of the ith set of equipment collected for the nth time; The material weight pre-storage library is as follows: In the formula, Weight (j) is the weight of the material collected for the rth set of equipment for the nth time; n Weight (j) is the weight of the material collected for the rth set of equipment for the nth time; The exhaust gas collection pre-storage library is as follows: In the formula, Gas_all(m, n) is the total amount of exhaust gas collected for the mth set of equipment for the nth time; n Gas_all(m, n) is the total amount of exhaust gas collected for the mth set of equipment for the nth time; The carbon emission pre-storage library is as follows: wherein, C_Dm,n is the carbon emission of the mth set of equipment in the nth collection; G_Dc is the carbon emission of the electric energy and the material; n C_Dtot,n is the total carbon emission in the nth collection; C_Dden is the carbon emission density; C_V is the volume, total exhaust gas quantity; The VOCs gas pre-storage library is as follows: wherein is the VOCs gas density; is the VOCs gas quantity; is the exhaust gas quantity; the pre-storage of particulate pollutants is as follows: wherein is the particle contamination density; is the particle contamination emission. The temperature pre-storage library is as follows: T x＝1 (k) = [t(l), t(2), t(3),..., t(k)] x＝1 (39) In the formula, T is temperature; yk is the index; k is the number of temperature samples; The pressure pre-storage library is as follows: S x＝1 (k) = [s(l), s(2), s(3),..., s(k)] x＝1 (42) In the formula, P is the pressure; The injection rate pre-storage library is as follows: V x＝1 (k) = [v(l), v(2), v(3),..., v(k)] x＝1 (45) In the formula, is the injection speed.

7. The modeling method of claim 1, wherein, In step 4), after obtaining the comprehensive emission prediction result, the comprehensive emission prediction result is compared with the emission warning threshold, and if the comprehensive emission prediction result is greater than the emission warning threshold, a warning is issued and a traceability search is performed to display the emission data and process parameters corresponding to the time period.

8. The modeling method of claim 1, wherein, The comprehensive emission prediction result includes the emission prediction results of carbon, VOCs gas, and particulate matter.

9. The modeling method of claim 1, wherein, In step 4), before inputting the current process parameters into the comprehensive emission prediction model, the process parameters are also denoised and dynamically range compressed.

10. An on-line dynamic evaluation system for the comprehensive emissions of a compounding injection molding process applying the method according to any one of claims 1 to 9, characterized in that, It comprises: a perception layer, an edge layer, and a cloud platform; the perception layer is used to collect process parameters; the edge layer denoises and dynamically range compresses the process parameters, and calls the comprehensive emission prediction model to process the denoised and dynamically range compressed process parameters to obtain the comprehensive emission prediction result; the cloud platform is used to train the comprehensive emission prediction model.

Images